Adaptive Schemes for Reducing Power Consumption of Mobile Data Handsets

This paper describes a threshold-based wake-up mechanism to reduce the battery power consumption of a mobile data handset. The threshold approach switches the system into the sleep mode when the memory queue for arriving packets is empty, and switches on the system when the number of packets in the memory queue is above a threshold value. We propose several adaptive schemes capable of dynamically selecting the threshold value for the threshold approach. An adaptive algorithm adjusts the threshold value based on a pre-defined packet-dropping probability, for which the switch-on rate is kept reasonably small to maintain the actual packet-dropping probability as close as the pre-defined value. Two strategies are used in the adaptive algorithm to adjust the threshold value: binary-division and fixed-amount. Two calculation strategies are considered to measure the packet-dropping probability: window-averaging and leaky-bucket integration (LBI). Our study indicates that the binary-division strategy outperforms the fixed-amount strategy in adjusting the threshold value. Furthermore, with proper setting, the LBI strategy outperforms the window-averaging strategy.